So tell me about a project that you worked on.

>> Sure let me tell you about some code we developed for

part of the third course in the specialization.

>> Okay.

>> So the project in that course is we want the learners to take some map data.

So some data and street data in the world.

So let's say we have some streets that look like this.

There's a street here and maybe there's this curvy street right there and

another street running here, maybe another street running there.

And the goal for the learners is that they want to find intersections in this map,

or take intersections in this map and then plot directions from one point to another.

So basically what they need to work with is these intersection points and

connections between them.

But this map data, as you can probably imagine, is real world and

it's extremely complex.

So there's a lot more than just these intersection points in this data.

In fact, in addition to these intersection points,

you actually get all these other little points along each of these lines.

So you've got all these yellow points here and ditto for all the other lines here.

>> So those dots trace the shape of the road.

>> Exactly, so these dots are necessary so that we can trace the shape of the road.

But in terms of finding routes,

direction planning, they're not really important, right?

>> Right. >> Because the only time you have a choice

is at an intersection.

>> Right.

>> So we could do some route planning on the full set of data here,

but the problem is that the graph that the users would have to work with,

the learners would work with, it would be very, very large.

So that would be kind of a problem.

So we wanted to be able to prune out all these yellow dots here so

that the learners didn't have to include them in their graph and allow the learners

to work just with these intersection points and the connections between them.

So, that was the technical problem that we had to deal with.

So, how did we deal with it?

[LAUGH] That's a good question.

So what we did was we first considered all of the data.

This is the data that comes back to us from the full map flaps.

And what we did was we represented all of the data you see here,

all the points in pink and in yellow, as a graph.